arXiv (Cornell University) · 2019 · 39 citations · 22 references
ComputingComputer ScienceComputational LinguisticsFeature EngineeringNatural Language ProcessingData SummarizationDeep LearningLanguage ModelingNatural Language UnderstandingText GenerationText SummarizationLanguage UnderstandingNeural Language ModelsNeural SummarizationNeural Text SummarizationSummarizationSummarization DatasetsTransformer Language ModelsTransformer Models
Large-scale learning of transformer language models has yielded improvements on a variety of natural language understanding tasks. Whether they can be effectively adapted for summarization, however, has been less explored, as the learned representations are less seamlessly integrated into existing neural text production architectures. In this work, we propose two solutions for efficiently adapting pretrained transformer language models as text summarizers: source embeddings and domain-adaptive training. We test these solutions on three abstractive summarization datasets, achieving new state of the art performance on two of them. Finally, we show that these improvements are achieved by producing more focused summaries with fewer superfluous and that performance improvements are more pronounced on more abstractive datasets.
22
Abstractive Text Summarization using Sequence-to-sequence RNNs and Beyond
Ramesh Nallapati, Bowen Zhou, Cícero dos Santos et al. · 2016 · 2.2K citations · Full text
Teaching Machines to Read and Comprehend
Karl Moritz Hermann, Tomáš Kočiský, Edward Grefenstette et al. · arXiv (Cornell University) · 2015 · 1.5K citations · Full text